?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
validation: delete obsolete dev_hold() check
[smatch.git] / smatch_kernel_user_data.c
blobcbe8943d10689b8136a1651f1e4e9a9f293ccee1
1 /*
2 * Copyright (C) 2011 Dan Carpenter.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version 2
7 * of the License, or (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, see http://www.gnu.org/copyleft/gpl.txt
16 */
17
18 /*
19 * There are a couple checks that try to see if a variable
20 * comes from the user. It would be better to unify them
21 * into one place. Also it we should follow the data down
22 * the call paths. Hence this file.
23 */
24
25 #include "smatch.h"
26 #include "smatch_slist.h"
27 #include "smatch_extra.h"
28
29 static int my_id;
30 static int my_call_id;
31
32 STATE(called);
33 static unsigned long func_gets_user_data;
34
35 static const char *kstr_funcs[] = {
36 "kstrtoull", "kstrtoll", "kstrtoul", "kstrtol", "kstrtouint",
37 "kstrtoint", "kstrtou64", "kstrtos64", "kstrtou32", "kstrtos32",
38 "kstrtou16", "kstrtos16", "kstrtou8", "kstrtos8", "kstrtoull_from_user"
39 "kstrtoll_from_user", "kstrtoul_from_user", "kstrtol_from_user",
40 "kstrtouint_from_user", "kstrtoint_from_user", "kstrtou16_from_user",
41 "kstrtos16_from_user", "kstrtou8_from_user", "kstrtos8_from_user",
42 "kstrtou64_from_user", "kstrtos64_from_user", "kstrtou32_from_user",
43 "kstrtos32_from_user",
44 };
45
46 static const char *returns_user_data[] = {
47 "simple_strtol", "simple_strtoll", "simple_strtoul", "simple_strtoull",
48 "kvm_register_read",
49 };
50
51 static struct stree *start_states;
52 static void save_start_states(struct statement *stmt)
53 {
54 start_states = clone_stree(__get_cur_stree());
55 }
56
57 static void free_start_states(void)
58 {
59 free_stree(&start_states);
60 }
61
62 static struct smatch_state *empty_state(struct sm_state *sm)
63 {
64 return alloc_estate_empty();
65 }
66
67 static struct smatch_state *new_state(struct symbol *type)
68 {
69 struct smatch_state *state;
70
71 if (!type || type_is_ptr(type))
72 return NULL;
73
74 state = alloc_estate_whole(type);
75 estate_set_new(state);
76 return state;
77 }
78
79 static void pre_merge_hook(struct sm_state *cur, struct sm_state *other)
80 {
81 struct smatch_state *user = cur->state;
82 struct smatch_state *extra;
83 struct smatch_state *state;
84 struct range_list *rl;
85
86 extra = __get_state(SMATCH_EXTRA, cur->name, cur->sym);
87 if (!extra)
88 return;
89 rl = rl_intersection(estate_rl(user), estate_rl(extra));
90 state = alloc_estate_rl(clone_rl(rl));
91 if (estate_capped(user) || is_capped_var_sym(cur->name, cur->sym))
92 estate_set_capped(state);
93 if (estate_treat_untagged(user))
94 estate_set_treat_untagged(state);
95 if (estates_equiv(state, cur->state))
96 return;
97 set_state(my_id, cur->name, cur->sym, state);
98 }
99
100 static void extra_nomod_hook(const char *name, struct symbol *sym, struct expression *expr, struct smatch_state *state)
101 {
102 struct smatch_state *user, *new;
103 struct range_list *rl;
104
105 user = __get_state(my_id, name, sym);
106 if (!user)
107 return;
108 rl = rl_intersection(estate_rl(user), estate_rl(state));
109 if (rl_equiv(rl, estate_rl(user)))
110 return;
111 new = alloc_estate_rl(rl);
112 if (estate_capped(user))
113 estate_set_capped(new);
114 if (estate_treat_untagged(user))
115 estate_set_treat_untagged(new);
116 set_state(my_id, name, sym, new);
117 }
118
119 static bool user_rl_known(struct expression *expr)
120 {
121 struct range_list *rl;
122 sval_t close_to_max;
123
124 if (!get_user_rl(expr, &rl))
125 return true;
126
127 close_to_max = sval_type_max(rl_type(rl));
128 close_to_max.value -= 100;
129
130 if (sval_cmp(rl_max(rl), close_to_max) >= 0)
131 return false;
132 return true;
133 }
134
135 static bool is_array_index_mask_nospec(struct expression *expr)
136 {
137 struct expression *orig;
138
139 orig = get_assigned_expr(expr);
140 if (!orig || orig->type != EXPR_CALL)
141 return false;
142 return sym_name_is("array_index_mask_nospec", orig->fn);
143 }
144
145 static bool binop_capped(struct expression *expr)
146 {
147 struct range_list *left_rl;
148 int comparison;
149 sval_t sval;
150
151 if (expr->op == '-' && get_user_rl(expr->left, &left_rl)) {
152 if (user_rl_capped(expr->left))
153 return true;
154 comparison = get_comparison(expr->left, expr->right);
155 if (comparison && show_special(comparison)[0] == '>')
156 return true;
157 return false;
158 }
159
160 if (expr->op == '&' || expr->op == '%') {
161 bool left_user, left_capped, right_user, right_capped;
162
163 if (!get_value(expr->right, &sval) && is_capped(expr->right))
164 return true;
165 if (is_array_index_mask_nospec(expr->right))
166 return true;
167 if (is_capped(expr->left))
168 return true;
169 left_user = is_user_rl(expr->left);
170 right_user = is_user_rl(expr->right);
171 if (!left_user && !right_user)
172 return true;
173
174 left_capped = user_rl_capped(expr->left);
175 right_capped = user_rl_capped(expr->right);
176
177 if (left_user && left_capped) {
178 if (!right_user)
179 return true;
180 if (right_user && right_capped)
181 return true;
182 return false;
183 }
184 if (right_user && right_capped) {
185 if (!left_user)
186 return true;
187 return false;
188 }
189 return false;
190 }
191
192 /*
193 * Generally "capped" means that we capped it to an unknown value.
194 * This is useful because if Smatch doesn't know what the value is then
195 * we have to trust that it is correct. But if we known cap value is
196 * 100 then we can check if 100 is correct and complain if it's wrong.
197 *
198 * So then the problem is with BINOP when we take a capped variable
199 * plus a user variable which is clamped to a known range (uncapped)
200 * the result should be capped.
201 */
202 if ((user_rl_capped(expr->left) || user_rl_known(expr->left)) &&
203 (user_rl_capped(expr->right) || user_rl_known(expr->right)))
204 return true;
205
206 return false;
207 }
208
209 bool user_rl_capped(struct expression *expr)
210 {
211 struct smatch_state *state;
212 struct range_list *rl;
213 sval_t sval;
214
215 expr = strip_expr(expr);
216 if (!expr)
217 return false;
218 if (get_value(expr, &sval))
219 return true;
220 if (expr->type == EXPR_BINOP)
221 return binop_capped(expr);
222 if ((expr->type == EXPR_PREOP || expr->type == EXPR_POSTOP) &&
223 (expr->op == SPECIAL_INCREMENT || expr->op == SPECIAL_DECREMENT))
224 return user_rl_capped(expr->unop);
225 state = get_state_expr(my_id, expr);
226 if (state)
227 return estate_capped(state);
228
229 if (!get_user_rl(expr, &rl)) {
230 /*
231 * The non user data parts of a binop are capped and
232 * also empty user rl states are capped.
233 */
234 return true;
235 }
236
237 if (rl_to_sval(rl, &sval))
238 return true;
239
240 return false; /* uncapped user data */
241 }
242
243 bool user_rl_treat_untagged(struct expression *expr)
244 {
245 struct smatch_state *state;
246 struct range_list *rl;
247 sval_t sval;
248
249 expr = strip_expr(expr);
250 if (!expr)
251 return false;
252 if (get_value(expr, &sval))
253 return true;
254
255 state = get_state_expr(my_id, expr);
256 if (state)
257 return estate_treat_untagged(state);
258
259 if (get_user_rl(expr, &rl))
260 return false; /* uncapped user data */
261
262 return true; /* not actually user data */
263 }
264
265 static void tag_inner_struct_members(struct expression *expr, struct symbol *member)
266 {
267 struct expression *edge_member;
268 struct symbol *base = get_real_base_type(member);
269 struct symbol *tmp;
270
271 if (member->ident)
272 expr = member_expression(expr, '.', member->ident);
273
274 FOR_EACH_PTR(base->symbol_list, tmp) {
275 struct symbol *type;
276
277 type = get_real_base_type(tmp);
278 if (!type)
279 continue;
280
281 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
282 tag_inner_struct_members(expr, tmp);
283 continue;
284 }
285
286 if (!tmp->ident)
287 continue;
288
289 edge_member = member_expression(expr, '.', tmp->ident);
290 set_state_expr(my_id, edge_member, new_state(type));
291 } END_FOR_EACH_PTR(tmp);
292 }
293
294 void __set_user_string(struct expression *expr);
295 static void tag_struct_members(struct symbol *type, struct expression *expr)
296 {
297 struct symbol *tmp;
298 struct expression *member;
299 int op = '*';
300
301 if (expr->type == EXPR_PREOP && expr->op == '&') {
302 expr = strip_expr(expr->unop);
303 op = '.';
304 }
305
306 FOR_EACH_PTR(type->symbol_list, tmp) {
307 type = get_real_base_type(tmp);
308 if (!type)
309 continue;
310
311 if (type->type == SYM_UNION || type->type == SYM_STRUCT) {
312 tag_inner_struct_members(expr, tmp);
313 continue;
314 }
315
316 if (!tmp->ident)
317 continue;
318
319 member = member_expression(expr, op, tmp->ident);
320 if (type->type == SYM_ARRAY) {
321 set_points_to_user_data(member);
322 } else {
323 set_state_expr(my_id, member, new_state(get_type(member)));
324 }
325 } END_FOR_EACH_PTR(tmp);
326 }
327
328 static void tag_base_type(struct expression *expr)
329 {
330 if (expr->type == EXPR_PREOP && expr->op == '&')
331 expr = strip_expr(expr->unop);
332 else
333 expr = deref_expression(expr);
334 set_state_expr(my_id, expr, new_state(get_type(expr)));
335 }
336
337 static void tag_as_user_data(struct expression *expr)
338 {
339 struct symbol *type;
340
341 expr = strip_expr(expr);
342
343 type = get_type(expr);
344 if (!type || type->type != SYM_PTR)
345 return;
346 type = get_real_base_type(type);
347 if (!type)
348 return;
349 if (type == &void_ctype) {
350 set_state_expr(my_id, deref_expression(expr), new_state(&ulong_ctype));
351 return;
352 }
353 if (type->type == SYM_BASETYPE) {
354 if (expr->type != EXPR_PREOP && expr->op != '&')
355 set_points_to_user_data(expr);
356 tag_base_type(expr);
357 return;
358 }
359 if (type->type == SYM_STRUCT || type->type == SYM_UNION) {
360 if (expr->type != EXPR_PREOP || expr->op != '&')
361 expr = deref_expression(expr);
362 else
363 set_state_expr(my_id, deref_expression(expr), new_state(&ulong_ctype));
364 tag_struct_members(type, expr);
365 }
366 }
367
368 static struct expression *ignore_param_set;
369 static void match_user_copy(const char *fn, struct expression *expr, void *_param)
370 {
371 int param = PTR_INT(_param);
372 struct expression *dest;
373
374 func_gets_user_data = true;
375 ignore_param_set = expr;
376
377 dest = get_argument_from_call_expr(expr->args, param);
378 dest = strip_expr(dest);
379 if (!dest)
380 return;
381 tag_as_user_data(dest);
382 }
383
384 static int is_dev_attr_name(struct expression *expr)
385 {
386 char *name;
387 int ret = 0;
388
389 name = expr_to_str(expr);
390 if (!name)
391 return 0;
392 if (strstr(name, "->attr.name"))
393 ret = 1;
394 free_string(name);
395 return ret;
396 }
397
398 static int ends_in_n(struct expression *expr)
399 {
400 struct string *str;
401
402 if (!expr)
403 return 0;
404 if (expr->type != EXPR_STRING || !expr->string)
405 return 0;
406
407 str = expr->string;
408 if (str->length < 3)
409 return 0;
410
411 if (str->data[str->length - 3] == '%' &&
412 str->data[str->length - 2] == 'n')
413 return 1;
414 return 0;
415 }
416
417 static void match_sscanf(const char *fn, struct expression *expr, void *unused)
418 {
419 struct expression *str, *format, *arg;
420 int i, last;
421
422 func_gets_user_data = true;
423
424 str = get_argument_from_call_expr(expr->args, 0);
425 if (is_dev_attr_name(str))
426 return;
427
428 format = get_argument_from_call_expr(expr->args, 1);
429 if (is_dev_attr_name(format))
430 return;
431
432 last = ptr_list_size((struct ptr_list *)expr->args) - 1;
433
434 i = -1;
435 FOR_EACH_PTR(expr->args, arg) {
436 i++;
437 if (i < 2)
438 continue;
439 if (i == last && ends_in_n(format))
440 continue;
441 tag_as_user_data(arg);
442 } END_FOR_EACH_PTR(arg);
443 }
444
445 static int get_rl_from_function(struct expression *expr, struct range_list **rl)
446 {
447 int i;
448
449 if (expr->type != EXPR_CALL || expr->fn->type != EXPR_SYMBOL ||
450 !expr->fn->symbol_name || !expr->fn->symbol_name->name)
451 return 0;
452
453 for (i = 0; i < ARRAY_SIZE(returns_user_data); i++) {
454 if (strcmp(expr->fn->symbol_name->name, returns_user_data[i]) == 0) {
455 *rl = alloc_whole_rl(get_type(expr));
456 return 1;
457 }
458 }
459 return 0;
460 }
461
462 static int comes_from_skb_data(struct expression *expr)
463 {
464 expr = strip_expr(expr);
465 if (!expr || expr->type != EXPR_PREOP || expr->op != '*')
466 return 0;
467
468 expr = strip_expr(expr->unop);
469 if (!expr)
470 return 0;
471 if (expr->type == EXPR_BINOP && expr->op == '+')
472 expr = strip_expr(expr->left);
473
474 return is_skb_data(expr);
475 }
476
477 static int handle_get_user(struct expression *expr)
478 {
479 char *name;
480 int ret = 0;
481
482 name = get_macro_name(expr->pos);
483 if (!name || strcmp(name, "get_user") != 0)
484 return 0;
485
486 name = expr_to_var(expr->right);
487 if (!name || (strcmp(name, "__val_gu") != 0 && strcmp(name, "__gu_val") != 0))
488 goto free;
489 set_state_expr(my_id, expr->left, new_state(get_type(expr->left)));
490 ret = 1;
491 free:
492 free_string(name);
493 return ret;
494 }
495
496 static bool state_is_new(struct expression *expr)
497 {
498 struct smatch_state *state;
499
500 state = get_state_expr(my_id, expr);
501 if (estate_new(state))
502 return true;
503
504 if (expr->type == EXPR_BINOP) {
505 if (state_is_new(expr->left))
506 return true;
507 if (state_is_new(expr->right))
508 return true;
509 }
510 return false;
511 }
512
513 static bool handle_op_assign(struct expression *expr)
514 {
515 struct expression *binop_expr;
516 struct smatch_state *state;
517 struct range_list *rl;
518
519 switch (expr->op) {
520 case SPECIAL_ADD_ASSIGN:
521 case SPECIAL_SUB_ASSIGN:
522 case SPECIAL_AND_ASSIGN:
523 case SPECIAL_MOD_ASSIGN:
524 case SPECIAL_SHL_ASSIGN:
525 case SPECIAL_SHR_ASSIGN:
526 case SPECIAL_OR_ASSIGN:
527 case SPECIAL_XOR_ASSIGN:
528 case SPECIAL_MUL_ASSIGN:
529 case SPECIAL_DIV_ASSIGN:
530 binop_expr = binop_expression(expr->left,
531 op_remove_assign(expr->op),
532 expr->right);
533 if (!get_user_rl(binop_expr, &rl))
534 return true;
535
536 rl = cast_rl(get_type(expr->left), rl);
537 state = alloc_estate_rl(rl);
538 if (expr->op == SPECIAL_AND_ASSIGN ||
539 expr->op == SPECIAL_MOD_ASSIGN ||
540 user_rl_capped(binop_expr))
541 estate_set_capped(state);
542 if (user_rl_treat_untagged(expr->left))
543 estate_set_treat_untagged(state);
544 if (state_is_new(binop_expr))
545 estate_set_new(state);
546 set_state_expr(my_id, expr->left, state);
547 return true;
548 }
549 return false;
550 }
551
552 static void match_assign(struct expression *expr)
553 {
554 struct symbol *left_type, *right_type;
555 struct range_list *rl;
556 static struct expression *handled;
557 struct smatch_state *state;
558 struct expression *faked;
559 bool is_capped = false;
560 bool is_new = false;
561
562 left_type = get_type(expr->left);
563 if (left_type == &void_ctype)
564 return;
565
566 faked = get_faked_expression();
567
568 /* FIXME: handle fake array assignments frob(&user_array[x]); */
569
570 if (is_fake_call(expr->right) && faked &&
571 faked->type == EXPR_ASSIGNMENT &&
572 points_to_user_data(faked->right)) {
573 if (is_skb_data(faked->right))
574 func_gets_user_data = true;
575 rl = alloc_whole_rl(get_type(expr->left));
576 is_new = true;
577 goto set;
578 }
579
580 if (faked && faked == handled)
581 return;
582 if (is_fake_call(expr->right))
583 goto clear_old_state;
584 if (handle_get_user(expr))
585 return;
586 if (points_to_user_data(expr->right) &&
587 is_struct_ptr(get_type(expr->left))) {
588 handled = expr;
589 // This should be handled by smatch_points_to_user_data.c
590 // set_points_to_user_data(expr->left);
591 }
592
593 if (handle_op_assign(expr))
594 return;
595 if (expr->op != '=')
596 goto clear_old_state;
597
598 /* Handled by DB code */
599 if (expr->right->type == EXPR_CALL || __in_fake_parameter_assign)
600 return;
601
602 if (!get_user_rl(expr->right, &rl))
603 goto clear_old_state;
604 is_capped = user_rl_capped(expr->right);
605 is_new = state_is_new(expr->right);
606
607 set:
608 if (type_is_ptr(left_type)) {
609 right_type = get_type(expr->right);
610 if (right_type && right_type->type == SYM_ARRAY)
611 set_points_to_user_data(expr->left);
612 return;
613 }
614
615 rl = cast_rl(left_type, rl);
616 state = alloc_estate_rl(rl);
617 if (is_new)
618 estate_set_new(state);
619 if (is_capped)
620 estate_set_capped(state);
621 if (user_rl_treat_untagged(expr->right))
622 estate_set_treat_untagged(state);
623
624 set_state_expr(my_id, expr->left, state);
625
626 return;
627
628 clear_old_state:
629
630 /*
631 * HACK ALERT!!! This should be at the start of the function. The
632 * the problem is that handling "pointer = array;" assignments is
633 * handled in this function instead of in kernel_points_to_user_data.c.
634 */
635 if (type_is_ptr(left_type))
636 return;
637
638 if (get_state_expr(my_id, expr->left))
639 set_state_expr(my_id, expr->left, alloc_estate_empty());
640 }
641
642 static void handle_eq_noteq(struct expression *expr)
643 {
644 struct smatch_state *left_orig, *right_orig;
645
646 left_orig = get_state_expr(my_id, expr->left);
647 right_orig = get_state_expr(my_id, expr->right);
648
649 if (!left_orig && !right_orig)
650 return;
651 if (left_orig && right_orig)
652 return;
653
654 if (left_orig) {
655 set_true_false_states_expr(my_id, expr->left,
656 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
657 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
658 } else {
659 set_true_false_states_expr(my_id, expr->right,
660 expr->op == SPECIAL_EQUAL ? alloc_estate_empty() : NULL,
661 expr->op == SPECIAL_EQUAL ? NULL : alloc_estate_empty());
662 }
663 }
664
665 static struct range_list *strip_negatives(struct range_list *rl)
666 {
667 sval_t min = rl_min(rl);
668 sval_t minus_one = { .type = rl_type(rl), .value = -1 };
669 sval_t over = { .type = rl_type(rl), .value = INT_MAX + 1ULL };
670 sval_t max = sval_type_max(rl_type(rl));
671
672 if (!rl)
673 return NULL;
674
675 if (type_unsigned(rl_type(rl)) && type_bits(rl_type(rl)) > 31)
676 return remove_range(rl, over, max);
677
678 return remove_range(rl, min, minus_one);
679 }
680
681 static void handle_compare(struct expression *expr)
682 {
683 struct expression *left, *right;
684 struct range_list *left_rl = NULL;
685 struct range_list *right_rl = NULL;
686 struct range_list *user_rl;
687 struct smatch_state *capped_state;
688 struct smatch_state *left_true = NULL;
689 struct smatch_state *left_false = NULL;
690 struct smatch_state *right_true = NULL;
691 struct smatch_state *right_false = NULL;
692 struct symbol *type;
693 sval_t sval;
694
695 left = strip_expr(expr->left);
696 right = strip_expr(expr->right);
697
698 while (left->type == EXPR_ASSIGNMENT)
699 left = strip_expr(left->left);
700
701 /*
702 * Conditions are mostly handled by smatch_extra.c, but there are some
703 * times where the exact values are not known so we can't do that.
704 *
705 * Normally, we might consider using smatch_capped.c to supliment smatch
706 * extra but that doesn't work when we merge unknown uncapped kernel
707 * data with unknown capped user data. The result is uncapped user
708 * data. We need to keep it separate and say that the user data is
709 * capped. In the past, I would have marked this as just regular
710 * kernel data (not user data) but we can't do that these days because
711 * we need to track user data for Spectre.
712 *
713 * The other situation which we have to handle is when we do have an
714 * int and we compare against an unknown unsigned kernel variable. In
715 * that situation we assume that the kernel data is less than INT_MAX.
716 * Otherwise then we get all sorts of array underflow false positives.
717 *
718 */
719
720 /* Handled in smatch_extra.c */
721 if (get_implied_value(left, &sval) ||
722 get_implied_value(right, &sval))
723 return;
724
725 get_user_rl(left, &left_rl);
726 get_user_rl(right, &right_rl);
727
728 /* nothing to do */
729 if (!left_rl && !right_rl)
730 return;
731 /* if both sides are user data that's not a good limit */
732 if (left_rl && right_rl)
733 return;
734
735 if (left_rl)
736 user_rl = left_rl;
737 else
738 user_rl = right_rl;
739
740 type = get_type(expr);
741 if (type_unsigned(type))
742 user_rl = strip_negatives(user_rl);
743 capped_state = alloc_estate_rl(user_rl);
744 estate_set_capped(capped_state);
745
746 switch (expr->op) {
747 case '<':
748 case SPECIAL_UNSIGNED_LT:
749 case SPECIAL_LTE:
750 case SPECIAL_UNSIGNED_LTE:
751 if (left_rl)
752 left_true = capped_state;
753 else
754 right_false = capped_state;
755 break;
756 case '>':
757 case SPECIAL_UNSIGNED_GT:
758 case SPECIAL_GTE:
759 case SPECIAL_UNSIGNED_GTE:
760 if (left_rl)
761 left_false = capped_state;
762 else
763 right_true = capped_state;
764 break;
765 }
766
767 set_true_false_states_expr(my_id, left, left_true, left_false);
768 set_true_false_states_expr(my_id, right, right_true, right_false);
769 }
770
771 static void match_condition(struct expression *expr)
772 {
773 if (expr->type != EXPR_COMPARE)
774 return;
775
776 if (expr->op == SPECIAL_EQUAL ||
777 expr->op == SPECIAL_NOTEQUAL) {
778 handle_eq_noteq(expr);
779 return;
780 }
781
782 handle_compare(expr);
783 }
784
785 static void match_returns_user_rl(const char *fn, struct expression *expr, void *unused)
786 {
787 func_gets_user_data = true;
788 }
789
790 static int get_user_macro_rl(struct expression *expr, struct range_list **rl)
791 {
792 struct expression *parent;
793 char *macro;
794
795 if (!expr)
796 return 0;
797
798 macro = get_macro_name(expr->pos);
799 if (!macro)
800 return 0;
801
802 /* handle ntohl(foo[i]) where "i" is trusted */
803 parent = expr_get_parent_expr(expr);
804 while (parent && parent->type != EXPR_BINOP)
805 parent = expr_get_parent_expr(parent);
806 if (parent && parent->type == EXPR_BINOP) {
807 char *parent_macro = get_macro_name(parent->pos);
808
809 if (parent_macro && strcmp(macro, parent_macro) == 0)
810 return 0;
811 }
812
813 if (strcmp(macro, "ntohl") == 0) {
814 *rl = alloc_whole_rl(&uint_ctype);
815 return 1;
816 }
817 if (strcmp(macro, "ntohs") == 0) {
818 *rl = alloc_whole_rl(&ushort_ctype);
819 return 1;
820 }
821 return 0;
822 }
823
824 static int has_user_data(struct symbol *sym)
825 {
826 struct sm_state *tmp;
827
828 FOR_EACH_MY_SM(my_id, __get_cur_stree(), tmp) {
829 if (tmp->sym == sym)
830 return 1;
831 } END_FOR_EACH_SM(tmp);
832 return 0;
833 }
834
835 bool we_pass_user_data(struct expression *call)
836 {
837 struct expression *arg;
838 struct symbol *sym;
839
840 FOR_EACH_PTR(call->args, arg) {
841 if (points_to_user_data(arg))
842 return true;
843 sym = expr_to_sym(arg);
844 if (!sym)
845 continue;
846 if (has_user_data(sym))
847 return true;
848 } END_FOR_EACH_PTR(arg);
849
850 return false;
851 }
852
853 static int db_returned_user_rl(struct expression *call, struct range_list **rl)
854 {
855 struct smatch_state *state;
856 char buf[48];
857
858 if (is_fake_call(call))
859 return 0;
860 snprintf(buf, sizeof(buf), "return %p", call);
861 state = get_state(my_id, buf, NULL);
862 if (!state || !estate_rl(state))
863 return 0;
864 *rl = estate_rl(state);
865 return 1;
866 }
867
868 struct stree *get_user_stree(void)
869 {
870 return get_all_states_stree(my_id);
871 }
872
873 static int user_data_flag;
874 static int no_user_data_flag;
875 struct range_list *var_user_rl(struct expression *expr)
876 {
877 struct smatch_state *state;
878 struct range_list *rl;
879 struct range_list *absolute_rl;
880
881 if (expr->type == EXPR_PREOP && expr->op == '&') {
882 no_user_data_flag = 1;
883 return NULL;
884 }
885
886 if (expr->type == EXPR_BINOP && expr->op == '%') {
887 struct range_list *left, *right;
888
889 if (!get_user_rl(expr->right, &right))
890 return NULL;
891 get_absolute_rl(expr->left, &left);
892 rl = rl_binop(left, '%', right);
893 goto found;
894 }
895
896 if (expr->type == EXPR_BINOP && expr->op == '/') {
897 struct range_list *left = NULL;
898 struct range_list *right = NULL;
899 struct range_list *abs_right;
900
901 /*
902 * The specific bug I'm dealing with is:
903 *
904 * foo = capped_user / unknown;
905 *
906 * Instead of just saying foo is now entirely user_rl we should
907 * probably say instead that it is not at all user data.
908 *
909 */
910
911 get_user_rl(expr->left, &left);
912 get_user_rl(expr->right, &right);
913 get_absolute_rl(expr->right, &abs_right);
914
915 if (left && !right) {
916 rl = rl_binop(left, '/', abs_right);
917 if (sval_cmp(rl_max(left), rl_max(rl)) < 0)
918 no_user_data_flag = 1;
919 }
920
921 return NULL;
922 }
923
924 if (get_rl_from_function(expr, &rl))
925 goto found;
926
927 if (get_user_macro_rl(expr, &rl))
928 goto found;
929
930 if (comes_from_skb_data(expr)) {
931 rl = alloc_whole_rl(get_type(expr));
932 goto found;
933 }
934
935 state = get_state_expr(my_id, expr);
936 if (state && estate_rl(state)) {
937 rl = estate_rl(state);
938 goto found;
939 }
940
941 if (expr->type == EXPR_CALL && db_returned_user_rl(expr, &rl))
942 goto found;
943
944 if (expr->type == EXPR_PREOP && expr->op == '*' &&
945 points_to_user_data(expr->unop)) {
946 rl = var_to_absolute_rl(expr);
947 goto found;
948 }
949
950 if (is_array(expr)) {
951 struct expression *array = get_array_base(expr);
952
953 if (!get_state_expr(my_id, array)) {
954 no_user_data_flag = 1;
955 return NULL;
956 }
957 }
958
959 return NULL;
960 found:
961 user_data_flag = 1;
962 absolute_rl = var_to_absolute_rl(expr);
963 return clone_rl(rl_intersection(rl, absolute_rl));
964 }
965
966 static bool is_ptr_subtract(struct expression *expr)
967 {
968 expr = strip_expr(expr);
969 if (!expr)
970 return false;
971 if (expr->type == EXPR_BINOP && expr->op == '-' &&
972 type_is_ptr(get_type(expr->left))) {
973 return true;
974 }
975 return false;
976 }
977
978 int get_user_rl(struct expression *expr, struct range_list **rl)
979 {
980 if (is_ptr_subtract(expr))
981 return 0;
982
983 user_data_flag = 0;
984 no_user_data_flag = 0;
985 custom_get_absolute_rl(expr, &var_user_rl, rl);
986 if (!user_data_flag || no_user_data_flag)
987 *rl = NULL;
988
989 return !!*rl;
990 }
991
992 int is_user_rl(struct expression *expr)
993 {
994 struct range_list *tmp;
995
996 return get_user_rl(expr, &tmp) && tmp;
997 }
998
999 int get_user_rl_var_sym(const char *name, struct symbol *sym, struct range_list **rl)
1000 {
1001 struct smatch_state *state;
1002
1003 state = get_state(my_id, name, sym);
1004 if (state && estate_rl(state)) {
1005 *rl = estate_rl(state);
1006 return 1;
1007 }
1008 return 0;
1009 }
1010
1011 static void return_info_callback(int return_id, char *return_ranges,
1012 struct expression *returned_expr,
1013 int param,
1014 const char *printed_name,
1015 struct sm_state *sm)
1016 {
1017 struct smatch_state *extra;
1018 struct range_list *rl;
1019 char buf[64];
1020
1021 if (param >= 0) {
1022 if (strcmp(printed_name, "$") == 0)
1023 return;
1024 if (!param_was_set_var_sym(sm->name, sm->sym))
1025 return;
1026 }
1027 rl = estate_rl(sm->state);
1028 if (!rl)
1029 return;
1030 extra = get_state(SMATCH_EXTRA, sm->name, sm->sym);
1031 if (estate_rl(extra))
1032 rl = rl_intersection(estate_rl(sm->state), estate_rl(extra));
1033 if (!rl)
1034 return;
1035
1036 snprintf(buf, sizeof(buf), "%s%s%s",
1037 show_rl(rl),
1038 estate_capped(sm->state) ? "[c]" : "",
1039 estate_treat_untagged(sm->state) ? "[u]" : "");
1040 sql_insert_return_states(return_id, return_ranges,
1041 estate_new(sm->state) ? USER_DATA_SET : USER_DATA,
1042 param, printed_name, buf);
1043 }
1044
1045 static bool is_ignored_macro(struct position pos)
1046 {
1047 const char *macro;
1048
1049 macro = get_macro_name(pos);
1050 if (!macro)
1051 return false;
1052 if (strcmp(macro, "v4l2_subdev_call") == 0)
1053 return true;
1054 return false;
1055 }
1056
1057 static void caller_info_callback(struct expression *call, int param, char *printed_name, struct sm_state *sm)
1058 {
1059 struct smatch_state *state;
1060 struct range_list *rl;
1061 struct symbol *type;
1062 char buf[64];
1063
1064 if (is_ignored_macro(call->pos))
1065 return;
1066
1067 /*
1068 * Smatch uses a hack where if we get an unsigned long we say it's
1069 * both user data and it points to user data. But if we pass it to a
1070 * function which takes an int, then it's just user data. There's not
1071 * enough bytes for it to be a pointer.
1072 *
1073 */
1074 type = get_arg_type(call->fn, param);
1075 if (strcmp(printed_name, "$") != 0 && type && type_bits(type) < type_bits(&ptr_ctype))
1076 return;
1077
1078 if (strcmp(sm->state->name, "") == 0)
1079 return;
1080
1081 state = __get_state(SMATCH_EXTRA, sm->name, sm->sym);
1082 if (!state || !estate_rl(state))
1083 rl = estate_rl(sm->state);
1084 else
1085 rl = rl_intersection(estate_rl(sm->state), estate_rl(state));
1086
1087 if (!rl)
1088 return;
1089
1090 snprintf(buf, sizeof(buf), "%s%s%s", show_rl(rl),
1091 estate_capped(sm->state) ? "[c]" : "",
1092 estate_treat_untagged(sm->state) ? "[u]" : "");
1093 sql_insert_caller_info(call, USER_DATA, param, printed_name, buf);
1094 }
1095
1096 static void db_param_set(struct expression *expr, int param, char *key, char *value)
1097 {
1098 struct expression *arg;
1099 char *name;
1100 struct symbol *sym;
1101 struct smatch_state *state;
1102
1103 while (expr->type == EXPR_ASSIGNMENT)
1104 expr = strip_expr(expr->right);
1105 if (expr->type != EXPR_CALL)
1106 return;
1107 if (expr == ignore_param_set)
1108 return;
1109
1110 arg = get_argument_from_call_expr(expr->args, param);
1111 if (!arg)
1112 return;
1113 name = get_variable_from_key(arg, key, &sym);
1114 if (!name || !sym)
1115 goto free;
1116
1117 state = get_state(my_id, name, sym);
1118 if (!state)
1119 goto free;
1120
1121 set_state(my_id, name, sym, alloc_estate_empty());
1122 free:
1123 free_string(name);
1124 }
1125
1126 static bool param_data_capped(const char *value)
1127 {
1128 if (strstr(value, ",c") || strstr(value, "[c"))
1129 return true;
1130 return false;
1131 }
1132
1133 static bool param_data_treat_untagged(const char *value)
1134 {
1135 if (strstr(value, ",u") || strstr(value, "[u"))
1136 return true;
1137 return false;
1138 }
1139
1140 static void set_param_user_data(const char *name, struct symbol *sym, char *key, char *value)
1141 {
1142 struct expression *expr;
1143 struct range_list *rl = NULL;
1144 struct smatch_state *state;
1145 struct symbol *type;
1146 char *fullname;
1147
1148 expr = symbol_expression(sym);
1149 fullname = get_variable_from_key(expr, key, NULL);
1150 if (!fullname)
1151 return;
1152
1153 type = get_member_type_from_key(expr, key);
1154 if (type && type->type == SYM_STRUCT)
1155 return;
1156
1157 if (!type)
1158 return;
1159
1160 str_to_rl(type, value, &rl);
1161 rl = swap_mtag_seed(expr, rl);
1162 state = alloc_estate_rl(rl);
1163 if (param_data_capped(value) || is_capped(expr))
1164 estate_set_capped(state);
1165 if (param_data_treat_untagged(value) || sym->ctype.as == 5)
1166 estate_set_treat_untagged(state);
1167 set_state(my_id, fullname, sym, state);
1168 }
1169
1170 static void set_called(const char *name, struct symbol *sym, char *key, char *value)
1171 {
1172 set_state(my_call_id, "this_function", NULL, &called);
1173 }
1174
1175 static void match_syscall_definition(struct symbol *sym)
1176 {
1177 struct symbol *arg;
1178 char *macro;
1179 char *name;
1180 int is_syscall = 0;
1181
1182 macro = get_macro_name(sym->pos);
1183 if (macro &&
1184 (strncmp("SYSCALL_DEFINE", macro, strlen("SYSCALL_DEFINE")) == 0 ||
1185 strncmp("COMPAT_SYSCALL_DEFINE", macro, strlen("COMPAT_SYSCALL_DEFINE")) == 0))
1186 is_syscall = 1;
1187
1188 name = get_function();
1189 if (!option_no_db && get_state(my_call_id, "this_function", NULL) != &called) {
1190 if (name && strncmp(name, "sys_", 4) == 0)
1191 is_syscall = 1;
1192 }
1193
1194 if (name && strncmp(name, "compat_sys_", 11) == 0)
1195 is_syscall = 1;
1196
1197 if (!is_syscall)
1198 return;
1199
1200 FOR_EACH_PTR(sym->ctype.base_type->arguments, arg) {
1201 set_state(my_id, arg->ident->name, arg, alloc_estate_whole(get_real_base_type(arg)));
1202 } END_FOR_EACH_PTR(arg);
1203 }
1204
1205 #define OLD 0
1206 #define NEW 1
1207
1208 static void store_user_data_return(struct expression *expr, char *key, char *value, bool is_new)
1209 {
1210 struct smatch_state *state;
1211 struct range_list *rl;
1212 struct symbol *type;
1213 char buf[48];
1214
1215 if (key[0] != '$')
1216 return;
1217
1218 type = get_type(expr);
1219 snprintf(buf, sizeof(buf), "return %p%s", expr, key + 1);
1220 call_results_to_rl(expr, type, value, &rl);
1221
1222 state = alloc_estate_rl(rl);
1223 if (is_new)
1224 estate_set_new(state);
1225
1226 set_state(my_id, buf, NULL, state);
1227 }
1228
1229 static void set_to_user_data(struct expression *expr, char *key, char *value, bool is_new)
1230 {
1231 struct smatch_state *state;
1232 char *name;
1233 struct symbol *sym;
1234 struct symbol *type;
1235 struct range_list *rl = NULL;
1236
1237 type = get_member_type_from_key(expr, key);
1238 name = get_variable_from_key(expr, key, &sym);
1239 if (!name || !sym)
1240 goto free;
1241
1242 call_results_to_rl(expr, type, value, &rl);
1243
1244 state = alloc_estate_rl(rl);
1245 if (param_data_capped(value))
1246 estate_set_capped(state);
1247 if (param_data_treat_untagged(value))
1248 estate_set_treat_untagged(state);
1249 if (is_new)
1250 estate_set_new(state);
1251 set_state(my_id, name, sym, state);
1252 free:
1253 free_string(name);
1254 }
1255
1256 static void returns_param_user_data(struct expression *expr, int param, char *key, char *value)
1257 {
1258 struct expression *arg;
1259 struct expression *call;
1260
1261 call = expr;
1262 while (call->type == EXPR_ASSIGNMENT)
1263 call = strip_expr(call->right);
1264 if (call->type != EXPR_CALL)
1265 return;
1266
1267 if (!we_pass_user_data(call))
1268 return;
1269
1270 if (param == -1) {
1271 if (expr->type != EXPR_ASSIGNMENT) {
1272 store_user_data_return(expr, key, value, OLD);
1273 return;
1274 }
1275 set_to_user_data(expr->left, key, value, OLD);
1276 return;
1277 }
1278
1279 arg = get_argument_from_call_expr(call->args, param);
1280 if (!arg)
1281 return;
1282 set_to_user_data(arg, key, value, OLD);
1283 }
1284
1285 static void returns_param_user_data_set(struct expression *expr, int param, char *key, char *value)
1286 {
1287 struct expression *arg;
1288
1289 func_gets_user_data = true;
1290
1291 if (param == -1) {
1292 if (expr->type != EXPR_ASSIGNMENT) {
1293 store_user_data_return(expr, key, value, NEW);
1294 return;
1295 }
1296 set_to_user_data(expr->left, key, value, NEW);
1297 return;
1298 }
1299
1300 while (expr->type == EXPR_ASSIGNMENT)
1301 expr = strip_expr(expr->right);
1302 if (expr->type != EXPR_CALL)
1303 return;
1304
1305 arg = get_argument_from_call_expr(expr->args, param);
1306 if (!arg)
1307 return;
1308 set_to_user_data(arg, key, value, NEW);
1309 }
1310
1311 static void returns_param_capped(struct expression *expr, int param, char *key, char *value)
1312 {
1313 struct smatch_state *state, *new;
1314 struct symbol *sym;
1315 char *name;
1316
1317 name = get_name_sym_from_key(expr, param, key, &sym);
1318 if (!name || !sym)
1319 goto free;
1320
1321 state = get_state(my_id, name, sym);
1322 if (!state || estate_capped(state))
1323 goto free;
1324
1325 new = clone_estate(state);
1326 estate_set_capped(new);
1327
1328 set_state(my_id, name, sym, new);
1329 free:
1330 free_string(name);
1331 }
1332
1333 static void match_function_def(struct symbol *sym)
1334 {
1335 if (is_user_data_fn(sym))
1336 func_gets_user_data = true;
1337 }
1338
1339 void register_kernel_user_data(int id)
1340 {
1341 int i;
1342
1343 my_id = id;
1344
1345 if (option_project != PROJ_KERNEL)
1346 return;
1347
1348 set_dynamic_states(my_id);
1349
1350 add_function_data(&func_gets_user_data);
1351 add_hook(&match_function_def, FUNC_DEF_HOOK);
1352
1353 add_hook(&save_start_states, AFTER_DEF_HOOK);
1354 add_hook(&free_start_states, AFTER_FUNC_HOOK);
1355 add_function_data((unsigned long *)&start_states);
1356
1357 add_unmatched_state_hook(my_id, &empty_state);
1358 add_extra_nomod_hook(&extra_nomod_hook);
1359 add_pre_merge_hook(my_id, &pre_merge_hook);
1360 add_merge_hook(my_id, &merge_estates);
1361
1362 add_function_hook("copy_from_user", &match_user_copy, INT_PTR(0));
1363 add_function_hook("__copy_from_user", &match_user_copy, INT_PTR(0));
1364 add_function_hook("memcpy_fromiovec", &match_user_copy, INT_PTR(0));
1365 for (i = 0; i < ARRAY_SIZE(kstr_funcs); i++)
1366 add_function_hook_late(kstr_funcs[i], &match_user_copy, INT_PTR(2));
1367 add_function_hook("usb_control_msg", &match_user_copy, INT_PTR(6));
1368 add_function_hook("kvm_read_guest_virt", &match_user_copy, INT_PTR(2));
1369
1370 for (i = 0; i < ARRAY_SIZE(returns_user_data); i++)
1371 add_function_hook(returns_user_data[i], &match_returns_user_rl, NULL);
1372
1373 add_function_hook("sscanf", &match_sscanf, NULL);
1374
1375 add_hook(&match_syscall_definition, AFTER_DEF_HOOK);
1376
1377 add_hook(&match_assign, ASSIGNMENT_HOOK);
1378 select_return_states_hook(PARAM_SET, &db_param_set);
1379 add_hook(&match_condition, CONDITION_HOOK);
1380
1381 add_caller_info_callback(my_id, caller_info_callback);
1382 add_return_info_callback(my_id, return_info_callback);
1383 select_caller_info_hook(set_param_user_data, USER_DATA);
1384 select_return_states_hook(USER_DATA, &returns_param_user_data);
1385 select_return_states_hook(USER_DATA_SET, &returns_param_user_data_set);
1386 select_return_states_hook(CAPPED_DATA, &returns_param_capped);
1387 }
1388
1389 void register_kernel_user_data2(int id)
1390 {
1391 my_call_id = id;
1392
1393 if (option_project != PROJ_KERNEL)
1394 return;
1395 select_caller_info_hook(set_called, INTERNAL);
1396 }
1397
Static analysis for C